How to fix heavy aliasing?

[cork_OS]

Hello.
Although moderate aliasing often can be successfully eliminated with QTGMC(InputType=2) or MAA2(), I have no idea how to fix or reduce heavy aliasing like this:
Sample1: http://www.mediafire.com/file/9pdclr068kpjsl9/aliasing1.wmv
https://s6.postimg.org/tivlnbxnh/aliasing1.png (https://postimg.org/image/tivlnbxnh/)
Sample2: http://www.mediafire.com/file/aai0sdmjcfgp3qo/aliasing2.mp4
https://s6.postimg.org/gsrda8pp9/aliasing2.png (https://postimg.org/image/gsrda8pp9/)

Any suggestions? Thanks in advance.

[Mounir]

try this:

main=last
main
NNEDI3(field=-2)
Merge(SelectEven(),SelectOdd())
filtered=last
main
mt_edge().mt_expand
mymask=last
mt_merge(main, filtered, mymask)


https://img4.hostingpics.net/pics/6213143101.jpg

[LemMotlow]

How were these videos deinterlaced? Why?

[cork_OS]

try this:
Thanks, your script is more effective for first sample. But for second sample it's not enough, unfortunately.

How were these videos deinterlaced? Why?
Actually, I don't know. Say, first video looks like ancient digitized VHS footage, but second is pretty fresh interview taken from local TV site (https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=ru&ie=UTF-8&u=http://www.tvr.by/events/proekty-atn/aktualnoe-intervyu/).
Although it's already 2017, there are deinterlacing artifacts in TV records, at least in my region (more examples: 1 (https://postimg.org/image/hoxi0jv31/), 2 (https://postimg.org/image/mozjm8q3h/)).

[hello_hello]

I bought a Bluray like that once. Very disappointing. Here's a 7MB sample (https://forum.videohelp.com/attachments/32479-1436357723/jaggie%20sample.mkv) I uploaded to another forum. The screen shots in this post (https://forum.videohelp.com/threads/372468-Is-x265-ready-for-Primetime-Migrating-From-x264-to-x265/page3#post2399335) were the result of a debate I was having at the time regarding resolution, but the fourth one has the filtering applied to repair the aliasing a bit. I think it'd been resized incorrectly and the aliasing was "baked in", and impossible to fix completely (for me anyway) so in the end I threw as much of it away as possible.

I think this is a little better than the previous suggestion for your second sample, and QTGMC will clean up some of the compression artefacts too, but it's also dog slow by comparison.

SeparateFields()
SelectOdd() # or selecteven if it looks better
nnedi3(dh=true)
QTGMC()
SelectOdd()
Gradfun3() # optional dithering